Long-term Social, Economic and Fiscal Effects of Immigration into the EU: The Role of the Integration Policy

d’Artis Kancs

Patrizio Lecca

2017

JRC Working Papers in Economics and Finance, 2017/4

This publication is a Technical report by the Joint Research Centre (JRC), the European Commission’s science

and knowledge service. It aims to provide evidence-based scientific support to the European policymaking

process. The scientific output expressed does not imply a policy position of the European Commission. Neither

the European Commission nor any person acting on behalf of the Commission is responsible for the use that

might be made of this publication.

Contact information

Name: Patrizio Lecca

Address: Edificio Expo, Calle Inca Garcilaso, 3, 41092 Sevilla, Spain Email: patrizio.lecca@ec.europa.eu

Tel.: +34 9544-88756

JRC Science Hub

https://ec.europa.eu/jrc

JRC107441

PDF ISBN 978-92-79-67441-9 ISSN 2467-2203 doi:10.2760/999095

Luxembourg: Publications Office of the European Union, 2017

© European Union, 2017

The reuse of the document is authorised, provided the source is acknowledged and the original meaning or

message of the texts are not distorted. The European Commission shall not be held liable for any consequences

stemming from the reuse.

How to cite this report: Kancs, D. and Lecca, P., (2017) Long-term Social, Economic and Fiscal Effects of

Immigration into the EU: The Role of the Integration Policy, JRC Working Papers in Economics and

Finance, 2017/4, doi:10.2760/999095

All images © European Union 2017

Long-term Social, Economic and Fiscal Effects ofImmigration into the EU: The Role of the Integration PolicyI

d’Artis Kancsa,∗, Patrizio Leccab

aEuropean Commission, DG Joint Research Centre, Ispra, Italy.bEuropean Commission, DG Joint Research Centre, Sevilla, Spain

Abstract

The issues of the forced migration and integration of refugees in the EU societyand labour markets are high on the policy agenda. Apart from humanitarian aspects,a sustainable integration of accepted refugees is important also for social, economic,budgetary and other reasons. Although, the potential consequences of the refugeeacceptance are being often discussed, little scientific evidence has been provided forthe policy debate so far in the context of the current refugee crisis. The present studyattempts to shed light on the long-run social, economic and budgetary effects of therapidly increasing forced immigration into the EU by performing a scenario analysis ofalternative refugee integration scenarios. Our simulation results suggest that, althoughthe refugee integration (e.g. by the providing language and professional training) iscostly for the public budget, in the medium- to long-run, the social, economic and fiscalbenefits may significantly outweigh the short-run refugee integration costs. Dependingon the integration policy scenario and policy financing method, the annual long-run GDPeffect would be 0.2% to 1.4% above the baseline growth, and the full repayment of theintegration policy investment (positive net present value) would be achieved after 9 to 19years.

Keywords: Migration, refugees, social inclusion, labour market, integration policy,modelling, scenario analysis, macroeconomic model.JEL code: F22, J6, J11, J24.

IThe authors acknowledge helpful suggestions from Laurent Aujean, Peter Bosch, Jan in ’t Veld, RalfJacob, Julda Kielyte, Rainer Munz, Damiaan Persyn, Jorg Peschner, Giuseppe Piroli, Romanos Priftis, FilipTanay and Alessandra Zampieri, participants of the EASO conference EU and Global Asylum-RelatedMigration Research in Malta, as well as participants of seminars at the Durham University and EuropeanCommission. The authors are solely responsible for the content of the paper. The views expressed arepurely those of the authors and may not in any circumstances be regarded as stating an official position ofthe European Commission.

∗Corresponding author, Competence Centre on Modelling.Email address: d'artis.kancs@ec.europa.eu (d’Artis Kancs)

1. Introduction

The increasing flows of forced migrants from civil war regions in the Middle East poserising challenges to EU societies and economies. In 2015 and 2016 the number of asylumseekers in the EU exceeded one million per year - for the first time since the World War II.These numerous inflows of refugees pose raising challenges to receiving countries. Thepresent study aims to assess the expected long-term social, economic and fiscal effects ofthe rapidly increasing forced civil-war refugee immigration into the EU and to accesswhat role an active policy can play in the refugee integration into the labour market.

When thinking about refugees, the humanitarian aspect is the most often invokedrationale. At the same time, however, forced civil war migrants pose also importantsocial and budgetary challenges, as well as may offer economic opportunities for EUeconomies. On the one hand, the social-beneficiary status quo of asylum applicants byproviding them with welfare benefits and the necessary access to education, languageand the social infrastructure may increase the budgetary costs of Member States in theshort-run. On the other hand, by integrating accepted asylum seekers into the EU labourmarkets may result not only in social, but in the long-run also in economic and budgetarygains. In addition, integrated refugees can play an important role, for example, in fillingvacancies with specific skill requirements, addressing Europe’s alarming demographicchallenges, improving the ratio of economically active to those who are inactive, a ratiothat is decreasing in many Member States, and boost jobs and growth in the EU.

A number of recent studies have investigated potential impacts of the migrant in-tegration into destination country labour markets. There is a general agreement inthe literature that, when immigrants bring skills that are short in supply in receivingcountries, there are good reasons to expect positive impacts on labour markets (Muenzet al. 2006). For example, Ottaviano and Peri (2012), show that general equilibriumeffects of immigration on receiving country’s labour markets depend strongly on thecomplementarity/substitutability between immigrants and foreign workers. Along thesame lines, also Peri (2016) confirms that the complementarity and substitutability be-tween immigrants and natives in employment, and the response of receiving economiesin terms of specialisation and technological choices, are important when assessing thegeneral equilibrium effects of immigration.

Most empirical studies, e.g. World Bank (2015), find that immigration can bringrelatively rapidly benefits to receiving country economies, depending on how quicklythe accepted refugees find jobs and to what extent their skills are complementary to the

1

existing workforce. Similarly, also Furlanetto and Robstad (2016) find that an exogenouspositive immigration shock lowers unemployment (even among native workers) andhas a measurable positive effect on public finances in the short-run. On the contrary,they do not find any support for those arguments recently used against immigration interms of native employment displacement effects and burden on public finances. Theseresults are consistent with finding of Jaumotte et al. (2016), according to which there aresignificant long-term benefits to immigration in terms of a higher GDP per capita forrecipient countries. Moreover, both high- and low-skill migrants contribute to the GDPper capita increase.

In contrast, other studies, e.g. Borjas (2014), provide evidence that there might be alsoadverse effects on destination country labour markets in terms of employment, wages andproductivity. In line with these arguments, Brucker and Jahn (2011) find that migrationreduces average wages and increases unemployment of the incumbent workforce inthe short-run, while it is neutral in the long-run. Further, OECD (2015) argues thatintegrating forced civil war migrants into receiving countries’ labour markets is morecostly than integrating economic migrants. According to OECD (2015), the integrationsuccess hinges crucially on integration policies, e.g. on providing the local languagetraining, the school education and professional training, and access to the socioculturalinfrastructure.

The existing literature, however, does not provide conclusive and robust answersto questions about the expected long-term effects of rapidly increasing asylum seekerinflows into the EU and the potential role of integration policies in the context of thecurrent refugee crisis. Among others, also because the concept of the skill complementar-ity/substitutability is less applicable to asylum seekers than to economic migrants. Thisstudy attempts to fill this knowledge gap by assessing the costs and benefits of the refugeeintegration into receiving EU Member States’ labour markets by identifying and account-ing for all key socio-economic and fiscal costs and benefits. Short- and medium-termintegration costs include, e.g. welfare benefits, the provision of the language training, theprofessional education to migrants and the school education to migrants’ children, accessto the sociocultural infrastructure, initial direct costs for the care, accommodation, and thecosts of providing social benefits for non-working refugees. In the short-term, a highergovernment spending on refugees may also have a positive impact on growth. However,given that any additional public spending needs to be financed through additional taxes,likely, the demand effect of the increased government consumption will not be the only

2

source of growth. Long-term economic benefits include, among others, integrated refugeecontributions to income taxes, payments to the welfare system (e.g. pension), as well asresolving labour market shortages in specific economic sectors and Member States.

In order to assess the impact of alternative refugee integration policies in the EU, weperform a scenario analysis with a Walrasian-type numerical general equilibrium model.The model, fully described in Mercenier et al. (2016), has a number of specificities thatsuit the purpose of the present study.1 First, given that the immigration of refugeesconcerns all EU Member States, the model captures the whole EU. Second, both regionaland sectoral dimensions are important for our questions, as refugee-related labour marketshocks will be very different across EU regions and sectors. Therefore, the model isimplemented at the regional level with a sectoral detail, regional and sectoral economiesbeing linked through the trade of goods and services, income flows, the labour and capitalmobility. Third, to capture the above-discussed skill complementarity/substitutabilityissue of migrants, the model distinguishes between three types of skills: low, mediumand high. Finally, the labour supply shock (accepted and integrated refugees enteringthe labour market) and the demand shock (additional expenditures of governmentand immigrants) will affect both the supply and demand sides of the labour market.The model captures not only the labour demand and labour supply, but also all keyinteractions between them. Labour market shocks are absorbed through three channels ofadjustment – (un)employment, wages and competitiveness – together with two exogenouslabour market mechanisms – education and migration.

The main contribution of the present study is to estimate the expected long-termsocial, economic and fiscal effects of the recently increasing inflows of forced civil-warrefugee migrants into the EU and to access what role an active policy can play in therefugee integration into the labour market. By complementing previous literature, ourstudy offers more nuanced insights regarding the regional and temporal distribution ofpotential impacts of the recent refugee crisis, the role of alternative financing methodsof government expenditure, and costs and consequences related to different policyengagement intensities in the integration of accepted refugees in EU labour markets. Thepresent study abstracts from economically-driven migration to the EU, as socio-economicimpacts of labour migration have been investigated considerably more and are relativelywell established in the existing literature.

1Previous version of the model has been used widely in several contexts. See, for example, Brandsma etal. (2014) and Brandsma et al. (2015).

3

The rest of the paper is structured as follows. Section 2 provides a brief backgroundabout the latest developments in terms of asylum seeker inflows into the EU, the refugeeintegration in the EU societies and labour markets, and presents an overview aboutthe relevant institutional framework. Section 3 provides a short overview about theunderlying simulation model, which will be used for simulating and assessing alternativerefugee integration policy scenarios. Section 4 outlines the main assumptions of the threeintegration policy scenarios (policy status quo, advanced and full), and of the baselinescenario. Section 5 presents and explains key simulation results. Section 6 presentsselected results from the sensitivity analysis. The final section concludes and draws anumber of policy conclusions from the present study.

2. Forced immigration in the EU: recent developments

2.1. Asylum seeker flows into the EU

EU Member States have been open to asylum seekers since a long time. Although,there have been several increases in the number of asylum seeker applications in the EUin the past (for example, a rather high number of asylum applications (672 thousand)were received in 1992, mainly from the former Yugoslavia), 2015 was the first time whenthe number of asylum applications in the EU exceeded one million in one year (seeFigure 1). Furthermore, according to IOM (2017), Europe might face high numbers ofasylum seekers/refugees also in the coming years.

Size and distribution of the forced migration. According to Figure 1, the annual number offirst time asylum applicants in the EU has historically been below half a million until 2013,with around 300 thousand per year on average between 2009 and 2016.2 The number offirst time asylum applicants has more than doubled in the EU from 563 thousand in 2014to almost 1.26 million in 2015. Also in 2016 the total number of asylum applicants in theEU significantly exceeded one million (Figure 1).

As regards destination countries, the largest increases in the number of first timeasylum applicants, expressed as a share of the total population, were recorded in Finland(around ten times higher in 2015 compared 2014), Hungary (around five times higher)and Austria (more than three times higher) (see Figure 2). In absolute numbers, the

2A first time applicant for international protection is a person who lodged an application for asylum forthe first time in a given EU Member State and therefore excludes repeat applicants in that Member Stateand so it reflects more accurately reflects the number of newly arrived persons applying for internationalprotection in the given Member State (Eurostat, 2016).

4

050

0000

1000

000

1500

000

asyl

um a

pplic

ants

2008 2010 2012 2014 2016year

Figure 1: Dynamics of asylum applications in the EU-28 Member States during 2008-2016, thousands.Source: Eurostat (online data code: migr_asyappctza).

inflow of first time asylum applicants in Germany increased from 173 thousand in 2014to 442 thousand in 2015 (Eurostat, 2016). Also Hungary, Sweden and Austria registeredvery large increases in first time asylum applications between 2014 and 2015. In contrast,most EU Member States in the Central and Eastern Europe, e.g. the Czech Republic,Croatia, Latvia, Lithuania, Slovenia and Romania, have registered rather few asylumapplicants both in 2014 and 2015. For migration modelling this implies that, in termsof the population share and the number of first time asylum applicants, differencesamong EU Member States are significant and hence we take them into account in thecomputations of country-specific costs and increases in the labour supply (section 4.1).

Age distribution. In terms of the age distribution, the vast majority of the first timeasylum seekers in the EU in 2015 were less than 35 years old (83%) (Eurostat, 2016);those in the age range 18-34 years accounted for slightly more than half (53%) of thetotal number of first time applicants, while almost one third (29%) were minors aged lessthan 18 years old. According to Eurostat (2016), more or less the same age distributionof asylum applicants was common across EU Member States, with the largest share

5

0 .5 1 1.5 2

United KingdomSweden

SpainSloveniaSlovakiaRomaniaPortugal

PolandNetherlands

MaltaLuxembourg

LithuaniaLatvia

ItalyIreland

HungaryGreece

GermanyFranceFinlandEstonia

DenmarkCzech Republic

CyprusCroatia

BulgariaBelgiumAustria

applicants as a share of population

Figure 2: Asylum applications in EU Member States in 2015 as a share of the incumbent population inpercent. Source: Eurostat (online data code: migr_asyappctza).

of applicants being those aged 18-34. The fact that, the majority of asylum seekers inthe EU are in the working age, will have consequences for the labour market, fiscaland budgetary impacts, and hence we take them into account in the computations ofcountry-specific costs and increases in the labour supply (section 4.1) Among others,they have the potential to strengthen the EU’s labour market and contribute to the EUMember States budget.

Acceptance and rejection. In 2015, more than half of all first instance asylum decisionsin the EU resulted in positive outcomes, granting a refugee or subsidiary protectionstatus, or an authorisation to stay for humanitarian reasons (Eurostat, 2016).3 For firstinstance decisions, around 75% of all positive decisions in the EU-28 resulted in grantsof refugee status in 2015, while for final decisions the share was somewhat lower, at69%. Also the share of successful applicants is important for the labour market, fiscaland budgetary impacts, and hence we take them into account, when calculating fiscalcosts per Member State for asylum seekers and for successful applicants, as well as the

3http://ec.europa.eu/eurostat/web/asylum-and-managed-migration/data/database

6

http://ec.europa.eu/eurostat/web/asylum-and-managed-migration/data/database

increase in the labour supply in receiving countries (section 4.1).The main contributions to the rapid increase in asylum seekers were growing numbers

of asylum applicants from civil war regions in Syria, Afghanistan and Iraq, 29%, 14%and 10%, respectively. Also Albania, Kosovo and Pakistan have significantly contributedto the increasing number of first time asylum applicants, 5%, 5% and 4%, respectively).Knowing the asylum seeker country of origin is important, as asylum seekers fromdifferent countries and world regions have systematic differences in language knowl-edge, education background and professional skills (Kielyte, 2016). Hence, there maybe differences in integration costs depending on where asylum seekers originate from,and differences in labour market, fiscal and budgetary impacts depending on the substi-tutability/complementarity of refugee skills (Ottaviano and Peri, 2012). In the presentstudy we follow Clemens and Pritchett (2016), and assume that migrants do not bringtheir economic productivity with them when moving to the destination country in the EU,and assume that during the training and learning process, accepted refugees assimilateskills and productivity of native-born.

2.2. Socio-economic situation of migrants in the EU

In virtually all EU Member States were accepted refugees reside, their standard ofliving is lower than that of the native-born population. According to OECD/EuropeanUnion (2015), the socio-economic situation of the migrant population in EU MemberStates is worse in several respects than that of native-born, particularly, in educationand employment. The European Agenda for the Integration of Third-Country Nationals(European Commission, 2011) summarises the most pressing challenges with respect tothe immigrant integration in the EU:

• Gaps in the educational achievement;

• Low employment levels of migrants, especially for migrant women;

• ‘Brain waste’ and high levels of ’over-qualification’;

• High risks of the social exclusion and poverty.

The existing evidence (e.g. OECD/European Union, 2015) suggests that third-countrynationals continue to face higher barriers in the education system than native-born. Inturn, low levels of education not only cause refugees’ social exclusion and poverty, theyalso create significant potential costs for Member State economies. Often, the latter aspect

7

has been neglected in the political debate, as not sufficient scientific evidence is available.When policy makers are thinking about investing more into the education of migrantsand their children, the social equity is the most often invoked rationale. At the same time,however, such investment may have a solid financial rationale as well. As noted by JamesHeckman, a Nobel laureate in economics:

“It is a rare public policy initiative that promotes fairness and social justiceand at the same time promotes productivity in the economy and in society atlarge. Investing in disadvantaged young children is such a policy.” (Heckman,2006)

According to OECD/European Union (2015), third-country nationals continue to facealso higher entry barriers in the labour market than native-born. Although, employmentis a core part of the integration process, migrants’ employment rates (62% at the EUaverage) remain below the average of host-country population in all EU Member States(OECD/European Union, 2015). Immigrant women tend to have particularly low employ-ment and activity rates in EU Member States. For the migration modelling, the relativelylow participation and employment rates of third-country nationals imply that they needto be accounted for when constructing alternative integration policy scenarios, as theyco-determine integration costs and labour market outcomes.

Further, immigrants with higher education degrees have greater difficulties of findinghigh-skill jobs than native-born. As a result, many immigrants are working in unskilledjobs despite having professional qualifications, they are overqualified for their jobs (whichoften are low-skill), or work in less favourable conditions in terms of wages and theemployment protection compared to native-born workers.

Finally, the income inequality is higher among immigrants than among native-born.Differences in wealth between third-country nationals and host country nationals aregreater than those between foreign-born and native-born. Also the poverty rate ofimmigrant households is considerably higher (around twice as high) than among hostcountry native households. In addition, according to OECD/European Union (2015),immigrants are more at the risk of the poverty or social exclusion compared to host-country nationals, even when they are in employment, and migrant children are exposedto a particularly high risk of poverty.

8

2.3. Policy framework of the immigrant integration in the EU

The EU Common Basic Principles for Immigrant Integration Policy adopted in 2004set out a common approach to immigrant integration in the EU.4 In an integrated Unionwithout internal borders, such as the Schengen area, the successful integration of third-country nationals should be a matter of a common interest to all EU Member States(European Commission, 2016). Therefore, the EU is increasingly supporting MemberStates in their immigrant integration policies. In 2011, the European Commission hasset out a European Agenda for the Integration of Third-Country Nationals,5 calling fora strengthened and coherent approach to integration across different policy areas andgovernment levels. The Agenda also highlights integration challenges, which EU MemberStates are facing. To address these challenges, it provides concrete recommendations andareas for action to support the integration of third-country nationals.

In terms of the migration policy, political signals from the EU have become particularlyclear and strong after the start of the refugee crisis at the end of 2014. Migration hasbecome one of the 10 priorities of the Juncker’s Commission. The European Agendaon Migration, which was proposed in May 2015,6 recognises migration both as anopportunity and challenge for the EU. It sets out medium to long-term priorities thatwill help EU Member States to manage migration challenges and, looking beyond crisesand emergencies, to capitalise opportunities. The Agenda has four pillars: 1) reducingthe incentives for irregular migration; 2) saving lives and securing external borders;3) strengthening the common asylum policy; 4) developing a new policy on the legalmigration. In line with Agenda’s priorities, actions and implementation packages havebeen discussed and proposed.

In November 2015, an Action Plan was signed at the Valletta Summit on Migration.7

At the Tripartite Social Summit of 16 March 2016, EU cross-industry social partnerspresented a joint statement on the refugee crisis,8 stressing the importance of the refugees’integration in training, employment and society in general. In its recent Communicationon the reform of the Common European Asylum System of 6 April 2016, the EuropeanCommission reassured that it is fully committed to achieve the objective of shaping

4Council conclusions of the Justice and Home Affairs Council of 5-6 June 2014.5COM (2011) 455 final of 20.7.2011.6COM(2015)240 final of 13.5.2012.7http://www.consilium.europa.eu/en/meetings/international-summit/2015/11/action_plan_

en_pdf/8https://www.etuc.org/sites/www.etuc.org/files/press-release/files/14.03.16_final_eco_

soc_partners_message_refugee_crisis.pdf

9

http://www.consilium.europa.eu/en/meetings/international-summit/2015/11/action_plan_en_pdf/http://www.consilium.europa.eu/en/meetings/international-summit/2015/11/action_plan_en_pdf/https://www.etuc.org/sites/www.etuc.org/files/press-release/files/14.03.16_final_eco_soc_partners_message_refugee_crisis.pdfhttps://www.etuc.org/sites/www.etuc.org/files/press-release/files/14.03.16_final_eco_soc_partners_message_refugee_crisis.pdf

an integrated, sustainable and holistic EU migration policy.9 On 12 April 2016, theEuropean Parliament adopted a Resolution calling, among others, for an early integrationand a full participation in the socio-economic life in host countries of all third-countrynationals, including refugees.10 In its recent Communication on the European Semester,the European Commission has clearly pointed out that:

“Economic and social policies will need to cater for the recent inflow ofmigrants and refugees, in particular to provide for their immediate needs andintegration in the labour market.”11

Regarding funding, under the current Multi-annual Financial Framework 2014-2020,765 million euro have been earmarked by EU Member States for the refugee integra-tion under the Asylum, Migration and Integration Fund (AMIF) national programmes(European Parliament, 2016). Further, the European Social Fund (ESF) with 86.4 billioneuro for 2014-2020 is gaining importance in the integration of refugees. For the initialphase of reception the Asylum, Migration and Integration Fund (AMIF) with 3.1 billioneuro has the mission to promote the refugee integration from a short-term perspective.The Fund for European Aid to the Most Deprived (FEAD) with 3.8 billion euro has themission to alleviate the refugee poverty and exclusion, it supports EU Member Statesin the integration of refugees within the scope of their budgets (European Parliament,2016). In addition, as part of the European Agenda for Migration, the Commission ispromoting a flexible and effective use of existing funds for refugees. As for the refugeeintegration into the vocational training, the European Commission is mobilising existingpolicy initiatives and programmes (European Alliance for Apprenticeships, EuropeanPact for Youth, Erasmus+, etc.) to promote the peer learning and sharing of promisingpractices on the refugee integration in the area of the vocational education and training.

3. Modelling framework

In this section, we briefly describe the main model’s features.12 The domestic economy(which corresponds to the EU) consists of 267 endogenous regions, which are includedinto EU Member States. The rest of the world is introduced in the model as an exogenous

9COM(2016) 197 final.10European Parliament Resolution (2015/2095(INI)).11COM(2016) 95 final.12The underlying model is fully described in Mercenier et al. (2016).

10

external sector. The economy is composed of different sectors (also called industries) inwhich firms operate under perfect competition.13 In each region-sector, a representativefirm produces a single variety, which is considered an imperfect substitute for the varietyproduced in other regions. Final goods are consumed by households, governments andinvestors (the form of capital goods), whilst firms consume intermediate inputs.

Trade between and within regions is costly, implying that the shipping of goodsentails transport costs assumed to be of the iceberg type. Transport costs are specific tosector and region pairs, they are based on the transport network model TRANSTOOLS,which considers different modes of transport and computes gneralised transportationcosts. These costs are positive within regions and asymmetric between regions.

Regional goods are produced by combining the value added (labour and capital) withdomestic and imported intermediates. Regional investments are determined by a simpleadjustment rule, according to which the additional level of investments is generated ineach region by the gap between the desired level of capital and the actual level of capital(Jorgenson and Stephenson, 1969, and Uzawa, 1969). Given the regional dimension of themodel, the demand for investment can be satisfied by domestic and external markets,without constraints. The capital stock in each region is updated period by period throughinvestments adjusted for depreciation.

To capture the above-discussed skill complementarity/substitutability issue of mi-grants, the model distinguishes between three skill levels of workers: low-skill, medium-skill and high-skill, which correspond to three levels of education: primary and lower-secondary education (ISCED 0-2), upper- secondary and post-secondary education(ISCED 3-4), and tertiary education (ISCED 5-6). The labour supply shock (accepted andintegrated refugees entering the labour market) and the demand shock (additional expen-ditures of government and immigrants) will affect both the supply and demand sides ofthe labour market. The model captures not only the labour demand and labour supply,but also all key interactions between them. On the labour demand side, the relative priceof labour together with the demand for final goods and services determine the labourdemand. Companies (and the public sector) rent labour services from households, forwhat they receive a wage income. On the labour supply side, each region is populatedby workers, who rent their differently skilled labour to firms.

For each skill level, the wage setting relationship is represented by a wage curve

13The simulation model has three market competition frameworks that can be activated: monopolisticcompetition, Cournot or Bertrand. In this study, we simplify the analysis by neglecting alternative pricesetting behaviours in favour of perfectly competitive markets.

11

020

4060

8010

0W

age

rate

.2 .4 .6 .8 1Labour force

labour supply

wage curve

labour demand

Figure 3: Labour market equilibrium for one selected region – wage curve, labour supply and labourdemand curves

(Blanchower and Oswald, 1995), whose implication is that lower levels of unemploymentincrease the workers’ bargaining power, thereby increasing real wages. A graphicalrepresentation of the labour market equilibrium is given in Figure 3 for one selectedregion and one representative skill, where an empirically estimated wage curve, thelabour supply curve and the labour demand curve are plotted. The differences betweenthe wage curve and labour supply determine the number of unemployed, therefore thewage rate is above the equilibrium market clearing wage rate. The relative position ofall three curves will be affected by the increasing stock of asylum seekers in the EU.Typically we would expect the wage curve and supply curve to move downwards whilethe labour demand would shift upward.

All shift and share parameters are calibrated to reproduce the base year data set,represented by the inter-regional Social Accounting Matrix. The model calibration processassumes regional economies to be initially in a steady-state equilibrium. This means thatthe capital stock is calibrated to allow depreciation to be fully covered by investments.Parameters related to the elasticities of substitution both on the consumer and producer

12

side are based on similar models or derived from the econometric literature. Typically,we assume a rather low elasticity of substitution in production (0.3), a relatively higherelasticity of substitutions in consumption (1.5) and a fairly high for trade between regions(2.0). The elasticity of substitution between different types of skills equate to 2.0. Theinterest rate (faced by producers, consumers and investors) is set to 0.04, the rate ofdepreciation to 0.15.

In the context of the present study, an important advantage of the adopted model isthat it is able to capture not only the direct effects of labour market adjustments, but alsoall induced inter-regional short- and long-run general equilibrium effects of exogenousshocks, such as, the refugee immigration or policy interventions, subsidies/taxes of wagefor low-, medium- and high-skill labour or changes in the labour supply and/or demand.Further, in addition to those labour market mechanisms that are modelled explicitly inthe model, many more peculiarities of regional labour markets are captured implicitly.For example, cross-regional differences in skills and hence productivity levels amongpeople with the same level of education are captured by calibrating the model to baseyear Labour Force Survey (LFS) data that are region- and skill-specific. Productivitydifferences between workers located in different regions together with migration costsbetween regions capture sizeable inter-regional wage differences of workers with thesame education level.

4. Scenario construction

In order to undertake a scenario analysis and assess the impact of alternative refugeeintegration policies in the EU, first, a baseline scenario is constructed and simulated (as wecannot observe it). In the context of the refugee immigration, the key assumption concernsthe exogenous population growth. In the baseline scenario, we use Eurostat’s regionalpopulation projections, which provide ‘what-if’ scenarios about the likely future size andstructure of populations, based on assumptions for fertility, mortality and migration.14

Second, alternative refugee integration (counterfactual) scenarios are constructed andsimulated. All key assumptions behind the refugee integration scenario construction aredetailed below.

Following the empirical evidence of Clemens and Pritchett (2016), we assume thatmigrants do not bring their economic productivity with them when moving to the

14Note that IMF (2016) assumes that the inflow of first-time asylum seekers would grow at 5 percentannually starting from 2019, which is higher than the baseline population growth in our study. In order tosee the impact of a higher baseline population growth, we perform sensitivity analysis.

13

destination country in the EU. Second, inspired by findings of Clemens and Pritchett(2016), we assume that during the training and learning process, accepted refugeesassimilate the skills and productivity of native-born. In modelling terms, these twoassumptions will imply that EU firms hiring/employing workers are blind with respectto native- / foreign-born, they cannot distinguish between integrated refugees andincumbent workers.

A further assumption that we make throughout this study is that we focus solelyon the forced migration triggered by civil-wars in the Middle East, while abstractingfrom other forms of migration, such as the economic migration, politically or naturaldisaster-driven migration. This implies that all simulation results reported in this studyrefer solely to the recently observed and predicted forced immigrant inflows from civilwar regions in the Middle East to the EU. This assumption also implies that the totalnumber of both regular and irregular migrants to the EU is larger in reality, implyingalso that the true socio-economic effects to the EU society and economy may exceed thosereported in the present study.

4.1. Assumptions about the forced migration to the EU15

Size and distribution of the forced migration. In line with Eurostat (2016), we assumethat 1.26 million first-time asylum seekers have entered the EU in 2015 (see section 2)and, following IMF (2016), we assume that 1.3 million will enter the EU in 2016 and2017. In line with the most recent empirical evidence (Eurostat 2106; and Kielyte 2016),we depart from IMF modelling choices by assuming that starting from 2018 the forcedmigration of refugees into the EU will each year decline by about one quarter, implying0.98 million first-time asylum seekers will arrive into the EU in 2018. Also in followingyears the annual inflow of asylum seekers into the EU will decrease by around onequarter annually, implying 0.73 million in 2019, 0.55 million in 2020, 0.41 million in2021, etc. In around 10 years, the current civil war and refugee crisis will have no moresignificant impact on the immigration of refugees into the EU, and the annual numbersof asylum seeker applications will have reached the long-run steady state, as assumedin the baseline scenario. Second, following IMF (2016), we assume that the geographicdistribution of refugees across EU Member States in future years will be the same as itwas in 2015.

15In order to facilitate the comparability of our results, most of our scenario assumptions about theforced migration of asylum seekers into the EU follow the modelling choices of IMF (2016). We departfrom IMF assumptions only when there is more recent/robust evidence available about refugee flows,distribution and/or characteristics in the EU, which we clearly indicate.

14

These two key assumptions about the size and distribution of forced immigrants aredriven by the fact that there is not sufficiently robust empirical evidence available formore nuanced and precise predictions about the size, dynamics and distribution of futurerefugee flows into the EU. Note, however, that these two assumptions are consistent withthe objective of the present study, which is to understand the impact of alternative refugeeintegration policies rather than to predict migration flows per se. In order to explore theimpact of a higher/lower inflow of forced civil war migrants, we run extensive sensitivityanalysis, selected results of which are presented in section 6.

Acceptance and rejection. Following Eurostat (2016), we assume that 60 percent ofasylum seeker applications are accepted, which is also in line with IMF (2016). Thisimplies that around 0.78 million additional refugees are annually accepted in the EUbetween 2016 and 2017. As in the IMF study, we assume that the rejected applicantsreceive financial support for one year and then they leave the EU. We further assume that,while staying in the EU, rejected applicants do not enter the EU labour market. Although,this assumption may not be entirely consistent with the real situation in some MemberStates, there is too little robust evidence available about the employment of rejectedasylum applicants in the EU to construct more nuanced policy scenarios. As above,in order to ensure the robustness of this assumption, we run an extensive sensitivityanalysis, e.g. by allowing the rejected asylum seekers to enter low-skill jobs.

Age distribution. Following IMF (2016) and Eurostat (2016),16 we assume that the shareof the working-age population (15–64 years old) among all accepted asylum seekers is 80percent.17 This assumption is based on the average share among all asylum applicants in2015 (see section 2). According to Eurostat (2016), 19.4 percent are children in the agegroup 0-13, and 0.6 percent are 65 or older. Based on these age distribution statistics, weassume that in the following years the working age population of accepted applicantswill increase by 0.92% annually (the job entry rate of young adults is higher than theretirement rate). Finally, we assume that the age distribution is equal between acceptedand rejected applicants.

Labour market rights. Following IMF (2016), we assume that gradually the acceptedapplicants will become eligible to work in the EU. However, due to lengthy workpermission procedures, labour market rights are not granted instantly. As in the IMF

16Online data code: migr_asyappctza.17Note that the IMF assumes that the share of the working-age population among all accepted asylum

seekers is slightly higher at 81 percent. In the present study we adjust the working age population estimatesof the IMF by latest EUROSTAT data.

15

study, we assume that accepted asylum applicants become eligible to work after one year,which is roughly in line with the Reception Conditions Directive.18 While acknowledgingthat in some Member States, for example, in Belgium and Italy, the access to labourmarkets usually is granted faster, the underlying simulation model cannot account formonthly or quarterly adjustments, as its minimum temporal frequency is 1 year. Further,we assume that, while not eligible to work, accepted asylum seekers continue to receivewelfare benefits.

4.2. Integration policy scenarios

Labour market integration. The labour market integration of accepted asylum seekersdepends, among others, on implemented refugee integration policies in EU MemberStates (OECD/European Union, 2015). Together with international migration policymakers and refugee integration experts in the European Commission, we have designedthree alternative refugee integration scenarios:

• Policy status quo scenario;

• Advanced integration scenario;

• Full integration scenario.

In order to simulate and assess the impact of alternative refugee integration policyoptions, first, integration costs and expected labour market outcomes are computed foreach scenario. In the policy status quo scenario, we assume the current level of thegovernment expenditure (per refugee) for the refugee integration into labour markets,and the resulting level of refugee participation and employment rates for low-, medium-and high-skill workers. In the advanced integration scenario, we assume an increase inthe current integration expenditures by two times, which will result in higher than thecurrent employment and participation rates of accepted refugees. In the full integrationscenario, we assume a significant increase in refugee integration expenditures to achievecomparable professional and language skills to natives, and the same participationand employment rates as the native population. Hence, differences in participationand employment rates among the three scenarios are due to higher spending on theintegration of accepted refugees. All three scenarios are compared with the baselinescenario.

18Directive 2013/33/EU of the European Parliament and of the Council.

16

Generic costs of asylum seekers.19 Part of fiscal costs to the public budget related tothe asylum seeker acceptance in EU Member States does not depend on the integrationpolicy. For example, during the application procedure, indispensable asylum seekerbenefits include accommodation costs, basic health care costs, in-kind benefits, cashbenefits, etc. In order to account for these costs, in all simulated policy scenarios weuse country-specific estimates on the public expenditure for refugees.20 For example,for Germany we use German Council of Economic Experts estimates on the public costof refugees: 800 EUR per asylum applicant plus 6600 EUR welfare benefits per year forsuccessful applicants. Note however that additional public administration costs e.g. forprocessing asylum seeker applications are not included in these calculations. In order toapproximate these additional costs in our simulations, we assume a fixed country-specificpublic administration cost based on the 2015 inflow of refugees, these costs not changewith the number of asylum seekers in the following years.

Scenario-specific costs of the refugee integration. Another part of fiscal costs to the publicbudget related to the refugee acceptance, education and integration into the labour marketdepends on integration policy efforts – achieving higher levels of refugee integration ismore costly for the public budget. Two types of integration costs are simulated in thepresent study: (i) language training costs; and (ii) upper secondary, professional and/ortertiary education costs. As noted by the European Commission (2016), learning thedestination country’s language is an important precondition for migrants to successfullystart their integration process into the host country’s society. In order to maximisethe learning effectiveness, language integration programmes should be provided at theearliest stage possible after the refugee arrival and the language learning should becombined with the learning of other skills and competencies or work experiences.

In order to account for differences in the investment in refugee integration betweenalternative policy scenarios, both language and professional education costs depend onthe integration intensity. As scenario names suggests, the lowest integration costs arein the policy status quo scenario, the highest in the full integration scenario. Languagetraining costs are collected from international language training institutions for eachEU Member State, they range from 153 EUR/person in Bulgaria to 3616 EUR/personin Sweden per successful asylum applicant. Professional training and school educationcosts of accepted refugees are assumed to be the same as for natives. We acknowledge

19Integration costs are calculated in real terms, 2015 being the base.20Note that this assumption is different from IMF (2016), where authors assume that fiscal costs comprise

an equal support for asylum applicants of 12000 EUR/person/year in all countries.

17

that, depending on the background of immigrants (skills, education, age, languageand IT knowledge), in reality these costs may be either lower or higher compared tonative-born. Moreover, as noted in section 2, there may be also important differencesamong refugees arriving from different migrant origin countries. However, there arenot sufficient data available about migrant characteristics to differentiate educationcosts between refugees and native-born, this is a promising area for a future research.Education cost data are extracted from the Eurostat domain educ_uoe_fine. Also schooleducation and professional training costs are Member State- and skill-specific, they rangefrom 1120 EUR/person/year in Bulgaria to 26410 EUR/person/year in Denmark. Totaleducation / training costs for each integration scenario and Member State are calculatedbased on the number of persons in education / training and the cost per person toachieve scenario-specific labour market outcomes (see below).

Scenario-specific labour market outcomes. In line with the empirical evidence (Eurostat2106; and Kielyte 2016), we assume that the efforts (and expenditures) of the refugee inte-gration policy determines labour market outcomes in terms of participation, employmentand the wage rate of refugees (which are determined in the model). Higher integrationcosts are associated with a higher participation, more skilled jobs, a higher employmentand wage rates. Following IMF (2016), we assume that, in the policy status quo scenario,the participation rate of accepted refugee seekers remains 10 percentage points lowerthan among natives, the unemployment rate remains 30 percentage points higher thanamong natives, and the majority of refugees enter low-skill jobs.21 Non-participating andunemployed refugees receive welfare benefits without contributing to the governmentrevenue.

In the advanced integration scenario, we assume that, after completing a languageand professional training, the participation rate of accepted refugee seekers is 5 per-centage points lower than among natives, and the unemployment rate is 15 percentagepoints higher than among natives. Also these labour market outcomes of the advancedintegration scenario are consistent with IMF (2016) assumptions about the labour marketintegration of accepted refugee seekers. Finally, in the full integration scenario, weassume that, after completing the language and professional training, the labour marketoutcomes in terms of participation, skilled jobs and employment rates of accepted refugeeseekers are comparable to those of natives in the host Member States.

21In IMF (2016) this scenario is referred to as ‘slow integration scenario’.

18

5. Simulation results

In this section, we describe simulations results of the three integration policy scenarios,as outlined in section 4.2. Policy shocks simulated in the model imply a combination oftwo demand-side shocks and one supply-side shock, shifting the three labour marketcurves (see section 3). On the demand side, we account for the increase in the currentgovernment expenditures related to the costs of the refugee integration into the labourmarket and the corresponding reduction in the disposable household income (due tolower transfers / higher taxes). In contrast to IMF (2016), who assumes that associatedfiscal costs are not offset by any fiscal measures, we account for the fact that all integrationpolicy costs are fully financed by EU Member States. In particular, they are funded by EUhouseholds in the form of a reduction in government transfers to the household incomeor an increase in the income tax rate. Although, the Stability and Growth Pact (SGP) hasa built-in flexibility that allows EU Member States to respond to the refugee crisis,22 e.g.through additional borrowing, we believe that assuming that integration policy costscan be financed through the government borrowing would bias the interpretation ofsimulation results. For example, more integration policy expenditures could be alwaysbetter than less, if associated policy costs are not borne by anyone in the EU economy. Afurther demand-side shock is given by the private consumption increase stemming fromrefugees themselves.

On the supply-side, the inflow of refugees increases the labour supply in EU MemberStates. As detailed in section 4.2, the increase in the labour force is Member State-, skill-and year-specific, and differs between the three integration policy scenarios. Given thatintegration policy costs and the projected labour force increase vary according to the levelof the refugee integration, it is likely to observe differences in the simulated economicimpact between the three alternative refugee integration policy scenarios.

5.1. Main results

Aggregated simulation results are reported in Figure 4, where we plot for the wholeEU the percentage change in the GDP from base line values for the three refugeeintegration scenarios (policy status quo, advanced integration scenario and full integrationscenario) for the years 2016-2040. The results reported in Figure 4 suggest that there

22According to the SGP, Member States can deviate from the adjustment path toward the Medium-TermObjective in case of an “unusual event outside the control of the member state which has a major impacton the financial position of the general government”.

19

0.5

11.

5Im

pact

on

GD

P, %

2015 2020 2025 2030 2035 2040year

GDP_full GDP_advancedGDP_status_quo

Figure 4: The simulated impact on the GDP under three alternative refugee integration scenarios, deviationsfrom baseline in percent. Source: Authors’ simulations. Notes: The bottom line represents the policy statusquo. Advanced and full integration scenarios are hypothetical scenarios simulating a significant increase inthe government expenditure on the refugee integration (see section 4.2 for scenario assumptions).

is a remarkable difference in terms of the economic impact between the three refugeeintegration scenarios. While all three scenarios share a similar dynamics over the timeperiod considered, differences in the GDP impact between them increases as economiesexpand. As expected, for the entire period of the simulated integration policy shock, thefull integration scenario records the highest boost in the GDP compared to the policystatus quo and advanced integration scenarios.

In the short-run (2016), the GDP impact under all three scenarios is positive andequate to 0.05%, 0.12% and 0.23% for the policy status quo, advanced and full integrationscenarios, respectively. As more integrated refugee workers enter the labour market, theEU economy expands continuously until 2040, when it reaches a new steady state abovethe baseline growth path of the GDP of 0.15%, 0.64% and 1.31% for the policy status quo,advanced and full integration scenarios, respectively. Figure 4 also suggests that the GDPimpact is rather small in the short-term, as the multiplier associated to the increase inthe government expenditure is partially or fully offset by the reduction in the disposable

20

household income. Furthermore, in few initial periods, the labour supply shock is toosmall to amplify effects on economic activities. However, as more workers enter thelabour market as a result of the refugee integration, the augmented labour supply isable to fill vacancies, put a downward pressure on wages and therefore generate a fall incommodity prices. The ultimate effect is then an improved competitiveness which booststhe economies of refugees-integrating EU Member States even further.

As next, we investigate whether and to what extent the costs associated with refugeeintegration policies are offset by benefits generated from the increase in workforce. Toidentify relative benefits, Figure 5 compares the costs of the policy integration expressedin terms of the government expenditure as a share of the base year GDP (bars) andbenefits expressed as percentage changes in the GDP (red line) off base year values.Notice that, in terms of units, in Figure 5 both refugee integration policy costs andbenefits are directly comparable. Here the main focus of our analysis is on the overallGDP effect for the whole EU for the period 2016-2040 where, for the sake of comparability,all three integration policy scenarios are reported side-by-side.

The simulation results reported in Figure 5 suggest that, in the first years and for allintegration levels, integration policy costs are higher than economic benefits. In particular,for the policy status quo integration scenario, only in the very long-run a break-evenpoint of positive net benefits is achieved. However, we should remind that neverthelessmultiplier effects are positive for all three refugee integration scenarios. For the other twointegration policy scenarios, the difference between costs and benefits is negative until2024. This implies that, depending on the integration policy scenario, in the first eightto nine years the fiscal costs of the refugee integration would be higher than economicbenefits. Hereafter, economic benefits exceed fiscal costs. The positive difference is largerfor the full integration scenario where, despite of higher integration costs, the augmentedworkforce generates even larger GDP expansionary effects.

By and large, from this analysis we can see that, if the EU economy would expandthrough the integration of refugees into the labour market, then economic benefits wouldbe eventually greater than fiscal costs. This is confirmed also by Figure 6, where we showthe net present value (NPV) associated to each integration scenario until 2040. The NPVis calculated as difference between the discounted present value of the GDP deviationfrom the base year values obtained from the results of simulations and the discountedpresent value of the government expenditure shock. The discount rate applied in thesecalculations is equal to 0.05.

21

Figure5:The

costofintegration

policiesas

ashare

ofthe

GD

P(bars)

andthe

simulated

impacton

theG

DP

(redline)

forthe

threeintegration

scenarios.Source:Authors’sim

ulations.Notes:The

firstintegration

scenariorepresents

thepolicy

statusquo.A

dvancedand

fullintegrationscenarios

arehypotheticalscenarios

simulating

asignificantincrease

inthe

governmentexpenditure

onrefugee

integration.Seesection

4.2for

scenarioassum

ptions.

22

−15

0000

−10

0000

−50

000

050

000

Net

Pre

sent

Val

ue, m

illio

n E

UR

2015 2020 2025 2030 2035 2040year

NPV_full NPV_advancedNPV_status_quo

Figure 6: Net present value of the refugee integration in the EU under the three integration scenarios.Source: Authors’ simulations. Notes: The first integration scenario represents the policy status quo.Advanced and full integration scenarios are hypothetical scenarios simulating a significant increase in thegovernment expenditure on refugee integration. See section 4.2 for scenario assumptions.

The NPV in this context gives us the size of the (discounted) multiplier effect in eachperiod. Note that, even if the NPV is negative, not necessarily the overall impact on theeconomy is negative. It could simply mean that the multiplier effect is less than 1. TheNPV should give an idea of the time in which the economy is able to fully absorb theexogenous demand shock.

The results plotted in Figure 6 suggest that the NPV is negative in the initial yearsbut then eventually becomes positive for all three refugee integration scenarios. It isalso interesting to see that, under the full integration scenario, the NPV is the lowestduring the period 2016-2025 but it becomes the largest after around ten years (2026).Hence, our results suggest that, in order to be able to gain from the full potential of therefugee integration in the EU labour markets, receiving countries should be prepared toaccept certain costs in the short-run. Moreover, even these short-term disadvantages areassociated with positive growth effects for the whole EU.

23

5.2. Alternative policy financing methods

Until now, we have assumed that the additional government revenue required forfinancing the integration of refugees was fully financed through imposing a lump sumtransfer of the household income. Economically, this is the most efficient way to createadditional government revenues, as it does not affect the marginal tax rate. However, thismight not be necessarily true in reality in EU Member States. Therefore, we perform afurther set of simulations in which the additional government revenue required to financethe cost of refugee integration policies is collected through an increase in the labourincome tax rate. In these simulations, we treat government expenditures as exogenousand vary tax rates so as to generate a corresponding increase in the government revenue.Therefore, the labour income tax rate has to increase until the additional tax revenue israised to cover the additional migration policy expenditure, while taking into accountendogenous changes in prices.

These simulations allow us to compare the case of financing the necessary additionalgovernment expenditures through a simple reduction of transfers from governments tohouseholds (as the case analysed above) with the case in which the costs of the refugeeintegration are fully financed by a corresponding increase in the labour income tax. Thetwo types of policy-financing options are rather different, as economic agents will adjustto the policy shock differently. A reduction in transfers simply affects the demand-sideof the economy by reducing the household consumption accordingly, while an increasein the labour income tax rate has an adverse supply-side effect. Therefore, it is likely topartially or fully offset the effects of the increase in the government expenditure, if theaugmented labour supply resulting from more refugee immigrants entering the labourmarket is not sufficient to counteract the adverse effects of the labour taxation.

Under default model assumptions, a long-run wage curve determines the real wageafter tax, therefore, an increase in the government expenditure fully financed throughtaxes in the absence of offsetting effects coming from the increasing labour supplywill lead to a negative overall impact in the short- and long-run.23 However, witha continuous integration of refugees into the EU labour market, we would expect adownward pressure on wages and, if the labour supply shock is large enough, we arelikely to observe a fall in real wages and therefore in commodity prices that should inturn raise competitiveness and boost the EU economy. In order to better illustrate the

23Results could change if for instance we impose fixed nominal wages. In these circumstances it is likelyto expect conventional Keynesian multiplier effects.

24

potential effects of a distortionary taxation, in this experiment we only focus on the fullintegration scenario, where the increase in the labour force is significantly larger than inthe other two scenarios.

Figure 7: The cost of integration policies as a share of the GDP (bars) and the simulated impact on theGDP (red lines) for the full integration scenario under two alternative financing methods. Source: Authors’simulations.

The simulations results for the two policy financing methods are reported in Figure 7.As in the preceding section, we report the required overall increase in the governmentexpenditure as a share of the EU GDP (grey bars) and simulated GDP effects (as a shareof the EU GDP) obtained by financing the integration policy via a lump sum transfer(red solid line) and via changes in the income tax (red dashed line). The simulationresults reported in Figure 7 suggest that under the non-distortionary tax, the economicimpact on the EU is positive since the beginning of the implementation of the integrationpolicy, whilst for the distortionary taxation the change in the GDP is negative in firstyears until 2021. Hereafter, as more refugee workers are integrated into the labourforce, the positive effect of an increase in the labour supply comes fully into play byboosting the growth of the EU economy. In the first 7 periods, the labour supply shockis still not sufficient to fully counteract the negative impact of the increase in taxes.

25

This is not a surprising result. In this time frame, the net labour income tax increasessignificantly in order to cover the significant expenditures necessary to operationalise theintegration of immigrants, whereas the number of immigrants integrating into the labourmarket is still not large enough to contrast the negative impact of taxation. However, assoon as the government expenditure declines (still increasing but at a decreasing rate),distortionary effects on wages are reduced, while we have a continuous and increasinginflow of refugees integrating into the labour force, in turn generating positive effects inthe economy.

Figure 8: Net present value under two alternative financing methods, full integration scenario. Source:Authors’ simulations.

Next, in Figure 8 we report the net present value for the two alternative policyfinancing methods. In both cases, the short- and medium-run NPV is negative. In thelong-run, however, a full repayment of the integration policy investment is achievedunder both policy financing methods. The simulation results reported in Figure 8 suggestthat, to achieve a full repayment of integration policy investments, it would be necessaryto wait until 2034, which implies 8 more years than under the lump sum policy financing.Moreover, it is interesting to see that, while the gap between the two alternative policy

26

financing methods is wider at the beginning of the policy shock, as the level of therefugee integration increases period by period, differences become smaller. Generally, weexpect the case of non-distortionary taxation to generate a larger expansion of the EUeconomy.

5.3. Regional impacts

Next, we look at the regional distribution of the simulated impacts of the refugeeimmigration to the EU, as given the uneven acceptance and integration of refugees acrossEU Member States (see Figure 2), we expect important heterogeneities also in the regionaldistribution of social, economic and fiscal impacts. In order to understand how differentEU regions might be affected by the recently recorded and projected refugee inflow, inFigure 9 we map the simulated long-run GDP effects for the policy status quo, (left panel),the advanced integration scenario (middle panel) and the full integration scenario (rightpanel).

In line with our expectation hypothesis, those countries and regions that acceptand integrate more refugees (measured by the share of the incumbent population) areexpected to benefit most from refugee immigrants: Finland, Sweden, Denmark, Austria.Germany. Second, Figure 9 confirms findings reported in Figure 5, according to whichthe full integration scenario is associated with higher GDP growth effects than the policystatus quo and the advanced integration scenarios.

Next, we explore the impact of alternative financing methods on the relative competi-tiveness of EU regions. As shown in section 5.2, the underlying mechanisms resultingfrom the application of a non-neutral financing policy are substantially different from aneutral financing option, where integration policy costs are fully financed through a lumpsum tax on the household income. When the model is framed to account for adversesupply side effects coming from taxation, in the short-run, a low level of the migrantintegration into labour markets is not enough to lower wages, therefore, workers becomecostlier and all prices in the economy rise, generating significant negative impacts onthe economic activity exacerbated by a loss in competitiveness. On the other hand, inthe long-run, a full integration of accepted asylum seekers occurs, implying that theincreased pool of available workers is large enough to put a downward pressure onwages stimulating in such a way the labour demand in all sectors. The positive impactof a raise in active workers entering the labour market is then able to counteract thenegative supply side effects of taxation increasing competitiveness in all EU regions thusboosting the economy even further.

27

(1.402,4.5](.421,1.402]

(.283,.421](.202,.283]

(.097,.202](.072,.097]

(.052,.072](.04,.052]

(.016,.04](.007,.016]

[.001,.007]

Impact on G

DP

− policy status quo scenario, %

(1.402,4.5](.421,1.402]

(.283,.421](.202,.283]

(.097,.202](.072,.097]

(.052,.072](.04,.052]

(.016,.04](.007,.016]

[.001,.007]

Impact on G

DP

− advanced integration scenario, %

(1.402,4.5](.421,1.402]

(.283,.421](.202,.283]

(.097,.202](.072,.097]

(.052,.072](.04,.052]

(.016,.04](.007,.016]

[.001,.007]

Impact on G

DP

− full integration scenario, %

Figure9:Sim

ulatedlong-run

impact

onthe

GD

Pund

erthe

policystatus

quo(left

panel),thead

vancedintegration

scenario(m

idd

lepanel)

andthe

fullintegrationscenario

(rightpanel).Source:A

uthors’simulations.

28

In order to identify sources of competitiveness effects at the regional level, we plotchanges in the regional GDP against changes in exports and compare both policy financ-ing methods for the short- and long-run, respectively. Red dots denote regional changesin the GDP (horizontal axes) and exports (vertical axes) under the distortionary financing,while grey dots denote regional variations for the same set of output variables under thelump-sum taxation for financing the integration policy. In both methods of financing, weare referring to the full integration scenario.

Figure 10: Impact on the relative competitiveness of regions: short-run. Notes: Red dots: distortionaryfinancing; grey dots: lump-sum taxation. Source: Authors’ simulations.

In the short-run (Figure 10), we observe a high negative correlation between theGDP and exports for the case of a non-distortionary taxation. In almost all regionspositive changes in the GDP are related to negative changes in exports (Figure 10), asthe government expenditure shock creates an upward pressure on prices resulting, in areduction in exports. The counteracting effect of the increase in active workers enteringthe labour market is still too small to revert the direction of the policy crowding outeffect. On the contrary, a positive correlation, although not as pronounced, is observedwhere the recycling method implies changes in the rate of the labour income tax. In

29

this case, negative changes in the regional GDP are associated to losses in the regionalcompetitiveness.

Figure 11: Impact on the relative competitiveness of regions: long-run. Notes: Red dots: distortionaryfinancing; grey dots: lump-sum taxation. Source: Authors’ simulations.

In the long-run, we observe a rather different situation (Figure 11). Under bothmethods of financing, positive changes in the GDP are related to positive competitivenesseffects (Figure 11). As explained above, the labour supply shock dominates over othereffects. It is worth noticing that regions above the fitting lines are those benefitingparticularly strongly from improved competitiveness effects, as for example severalregions in Hungary, Sweden and Austria, while other regions experience less thanexpected changes in export as those below the fitting line.

As expected, under the non-distortionary taxation, the effects on the GDP and exportsare lower than under the neutral financing, while both variables show the same correlationeffects in both types of simulations. We can identify a group of regions with comparableGDP effects, while the same group of regions experiencing heterogeneous competitivenesseffects. In order to identify competitiveness adjustment channels, we regress changes inexports on changes in the regional GDP and plot residuals in Figure 12. Residuals, that

30

Figure 12: Expected versus realised gains in the relative competitiveness of regions. Positive values - morethan expected; negative values - less than expected. Source: Authors’ simulations.

is, differences between the actual changes and the predicted changes in exports, providea measure of the relative competitiveness that regions achieve as result of the integrationof accepted refugee migrants into their labour markets. In Figure 12 we report results forthe distortionary taxation, which are considerably more interesting than for the neutraltaxation. When residuals are negative, gains in the relative competitiveness resultingfrom the integration of migrants in a particular region are smaller than what we haveexpected. In the opposite case, that is when residuals are positive, the gains in the relativecompetitiveness are larger than expected. This analysis is useful, because it neutralisesthe direct effect related to the labour supply shock.

According to simulation results reported in Figure 12, Swedish regions receive asignificant increase in the numbers of active workers joining the labour market, however,resulting relative competitiveness gains are smaller than we would have expected, giventhe long-run impacts on the GDP. This is not the case regions of Hungary and Austria thatreceive a comparably large labour supply shock. Deviations between the expected and

31

realised gains in the relative competitiveness are due to heterogeneities across regions inlabour markets, the structure of economies, trade patterns, etc., as observed in the baseyear data.

Finally, to further facilitate the understanding of mechanisms operating in the modelat the regional level, as next we plot and discuss in detail the percentage changes inthe GDP, exports and household consumption for one representative region – the Parisregion of France (FR10, NUTS2) – for the advanced integration scenario, which is shownin Figure 13.

Figure 13: The simulated impact on the GDP, exports and the household consumption for the Paris regionin France, deviations from the baseline in percent. Source: Authors’ simulations.

In the very short-run, which corresponds to the first period of simulations (2016), thehousehold consumption is constrained to fall in order to finance the integration policycost, while exports (interregional and international) are slightly below the base yearvalue (-0.06%). In this time frame, GDP multiplier effects are still positive, since theGDP is above base year values, though the magnitude of the impact is negligible. Thisoccurs because the positive supply-side shocks, coming from the additional supply ofintegrated refugee workers entering into the labour market, are not strong enough to

32

fill all vacancies in the economy and generate a sufficient fall in prices to improve thecompetitiveness of refugees-integrating EU Member States’ economies. However, as thelabour supply rises through the increased level of the refugee integration, competitivenesseffects come fully into operation and the household consumption starts to increase. Thisoccurs after around 9 years, after which the household consumption keeps continuouslyrising until settling at around 0.16% off base line values.

5.4. Comparison with previous studies

Comparing our results with previous studies, we note that the range of the estimatedGDP effects of immigration on receiving countries is rather wide in the migration litera-ture, implying that many factors, such as migrant skill complementarity/substitutabilityto the incumbent workforce skills, integration policies, local labour market conditions,likely will determine the net effect of immigration in the medium- to long-run.

Our simulation results are consistent with findings of Jaumotte et al. (2016), accordingto which there might be significant long-term benefits to immigration in terms of ahigher GDP per capita for receiving countries. Their estimates suggest that an increase inthe migration share by 1 percentage point can raise the destination country per capitaGDP by up to 2 percent in the long-run. Moreover, both high- and low-skill migrantscontribute to the GDP per capita increase in receiving countries. Their estimates of asizeable and positive impact of immigration on the GDP per capita also suggest that thetrue fiscal benefits from immigration could be larger than typically estimated, since staticestimates of net fiscal gains, which calculate the difference between immigrants’ taxes andsocial security contributions and their receipt of social security benefits and governmentservices, typically do not take into consideration the indirect effects of immigration onthe aggregate productivity of the receiving country economy. Jaumotte et al. (2016)conclude that the labour market integration of migrants is critical to secure GDP percapita gains and benefits for public finances, and a number of policies can increase thebenefits of migrant worker integration, including language training and active labourmarket policies targeted to the needs of migrants. Hence, the policy conclusions of thepresent study are fully supported by those of Jaumotte et al. (2016).

The results of the IMF (2016) study, analysing the economic impact of the recentrefugee inflow in the EU, suggest that in the short-run, the macroeconomic effect from therefugee inflow in the EU is likely to be modest in terms of the GDP growth, reflecting thefiscal expansion associated with the support to asylum seekers, as well as the expansionin the labour supply, as integrated refugees begin to enter the EU labour market. IMF

33

(2016) concludes that the impact of accepted refugees on medium and long-term growthdepends on how they will be integrated in the EU labour market. Policies can help openup the refugee’ path to the EU labour market: restrictions on taking up work duringthe asylum application phase should be minimised, and active labour market policiesspecifically targeted to the refugees strengthened in the EU. Again, the policy conclusionsof the present study seem to be in line with those of IMF (2016).

Similarly, also Furlanetto and Robstad (2016) find that an exogenous positive immi-gration shock lowers unemployment (even among native workers) and has a positiveeffect on public finances even in the short-run. The positive long-run effects are likelyto be even larger. On the contrary, they do not find any support for those argumentsrecently used against immigration in terms of native employment displacement effectsand burden on public finances.

Our simulation results are also broadly comparable to the results reported in a studyexamining the economic effects of administrative action on immigration on the USeconomy by the U.S. President’s Council of Economic Advisors (CEA, 2014). The CEAestimates are based on a bottom-up approach relying on the literature’s estimates of theeffects of high-skill immigration on TFP, hours supplied per workers, skill composition ofthe workforce, and capital intensity. According to (CEA, 2014) results, the semi-elasticityof output per worker to immigration is of a similar magnitude to the one we have foundin the present study for the policy status quo scenario.

Further, in a cross-sectional setting, Ortega and Peri (2014) find that a country with amigration share of 10 percentage points higher than another country would have twiceas high level of income in the long-run. These estimates are comparable to our results forthe full integration scenario reported in the present study.

As a general conclusion, our results seem to be broadly in line with those of previousstudies investigating possible impacts of immigration on destination country societyand economy. By complementing previous findings, our study offers more nuancedinsights regarding the regional and temporal distribution of potential impacts, the role ofalternative financing methods of the government expenditure, and the consequences ofthe policy engagement and costs regarding the integration of accepted refugees in EUlabour markets.

34

6. Robustness checks and sensitivity analysis

6.1. Robustness checks

The success of the refugee integration policy will largely depend on the level of thelabour market integration, which the EU will be able to achieve. For this reason, wehave constructed three alternative integration scenarios by altering critical assumptionsabout integration policy costs and expected labour market outcomes. As the process ofacceptance, training and all other tasks related to the integration of refugees improvesdue to the exchange of best practices between Member States, learning by integrating andinformation sharing, the level of refugee integration policy effectiveness and efficiency islikely to improve over time. Hence, we can expect certain policy learning effects to arise,if all procedures linked to the integration of refugees become a common and regularpractice in the EU.

In order to provide an indicative measure of the likely impact of the efficiencyimprovement of the investment associated to the integration of refugees into the labourmarket and the extent to which a learning effect can generate a faster increase in theeconomic growth, we make a hypothetical assumption of increasing the efficiency of therefugee integration into the EU labour market in the three integration scenarios. Theinitial assumptions on participation, skilled jobs and the employment rate are preserved.However, in this robustness check we assume that the number of integrated refugeesentering the labour market increases by 20%, while keeping the government expenditureinvariant.

The improvement in the public spending not only boosts the GDP growth in the long-run but, as shown in Figure 14, in addition, it is also expected to reduce the short-runcrowding out effects of financing the cost of integrating refugees through a distortionarytaxation. These findings suggest that the simulation results reported in previous sectionsmay represent a lower bound of real refugee integration benefits. The true benefits maybe even higher, depending on the policy efficiency and spillover effects.

6.2. Sensitivity analysis

Given that the objective of the present study is to simulate, assess and compareimpacts of alternative refugee integration policies, those assumptions that differ amongscenarios are particularly critical for the robustness of our simulation results. Therefore,all these assumptions are scrutinised extensively in the sensitivity analysis. In contrast,as we do not attempt to forecast migration flows as such – they are taken as given (and

35

Figure 14: The EU GDP impact of the improved efficiency of the full integration scenario under twoalternative financing methods. Source: Authors’ simulations.

hence constant across all scenarios) in the present study – assumptions, for example,about the causes and determinants of the forced migration of refugees into the EU areless critical for simulation results of the present study.

In this section, we present selected results of the sensitivity analysis for the threeintegration scenarios. In particular, as an example, we report the results of the sensitivityanalysis obtained by altering the values of behaviour parameters in the underlyingsimulation model, as they are among key determinants driving simulation results.

In this exercise, we decrease and increase model default elasticities of substitutionby 20%, respectively, and compare with the effects under our default assumptions. Themain purpose of this exercise is to evaluate the extent to which changes in the elasticityof substitution would alter the magnitude of the simulated impact reported in precedingsections. The sensitivity analysis results are reported in Figure 15, where percentagedeviations in the GDP for all three refugee integration scenarios are reported. For eachscenario we show the case where elasticities are reduced (lower), augmented (upper) andkept at their default (central) values.

36

Full integration

Advanced integration

Status quo

0.5

11.

5Im

pact

on

GD

P, %

2015 2020 2025 2030 2035 2040year

GDP_status_quo_lower GDP_status_quo_central GDP_status_quo_upper GDP_partial_lower

GDP_partial_central GDP_partial_upper GDP_full_lower GDP_full_central

GDP_full_upper

Figure 15: Sensitivity analysis under the three refugee integration scenarios, deviations from the baselinein percent. Source: Authors’ simulations. Notes: The first integration scenario represents the policy statusquo. Advanced and full integration scenarios are hypothetical scenarios simulating a significant increase inthe government expenditure on refugee integration. See section 4.2 for scenario assumptions.

The sensitivity analysis results reported in Figure 15 suggest that in all three refugeeintegration scenarios the short-run impact of altering the values of the elasticities ofsubstitution do not generate important and significant variations in simulation results.However, as the labour force and economies grow, the elasticities of substitutions come inmore prominently to play a role. It is interesting to see that, by lowering the values of theelasticities of substitution, the impact is lower on the contrary by increasing all elasticitiesthe GDP impact becomes larger. The main channel in this analysis is surely representedby the trade elasticity. As the EU economy becomes more competitive, when morerefugee workers are integrated into the labour force, a rise in the trade elasticity allowseconomies to better exploit comparative advantages of the improved competitiveness.

In addition to presented sensitivity analysis results, we have also performed a numberof other sensitivity analyses with respect to the size, dynamics and distribution of theforced immigration (not reported). Our sensitivity analysis results suggest that indeedthese assumptions are among the key factors driving simulation results. Hence, an

37

updated set of simulation results may be produced in future, when more recent dataabout the size, dynamics and distribution of the forced immigration in the EU becomeavailable.

7. Conclusions

The rising numbers of forced civil war migrants pose important social and fiscalchallenges as well as offer economic opportunities for EU Member State societies andeconomies. In the short-run, the social-beneficiary status quo of asylum applicants, byproviding welfare benefits and the necessary access to education, language and thesociocultural infrastructure, increase the budgetary costs of EU Member States. In themedium- to long-run, by integrating migrants into EU labour markets offer not onlyin social, but also in economic and budgetary gains. In order to assess how differentrefugee integration policies can affect labour market outcomes, the economic growth andthe government budget of EU Member States in the medium- to long-run, the presentstudy undertakes a scenario analysis with a general equilibrium model for the whole EUimplemented at the regional level.

Our simulation results suggest that, although the refugee integration (e.g. by provid-ing welfare benefits, language and professional training) is costly for the governmentbudget, in the medium- to long-run, socio-economic and fiscal benefits may significantlyoutweigh the associated refugee integration costs. Most importantly, in the medium- tolong-run, the net benefits of investment into refugee integration are higher, the higherare investments into refugee integration at the time of their arrival. Depending on theintegration policy scenario and policy financing method, the annual long-run GDP effectwould be 0.2% to 1.4% above the baseline growth, and the full repayment of the integra-tion policy investment (positive the net present value) would be achieved after 9 to 19years. Hence, our study confirms that sustainably integrated refugees have the potentialto play an important role in addressing Europe’s alarming demographic trends, fillingvacancies with specific skill requirements, improving the ratio of economically active tothose who are inactive, a ratio that is falling in many Member States, and boosting jobsand growth in the EU.

Based on our simulation results, we can draw a number of policy recommendations.First, an effective and sustainable integration policy is costly, it requires sufficient political,social, and financial resources and investments which, according to our simulation results,will pay off for receiving countries in the medium- to long-run (after 9 years). Second, the

38

integration of